Co.) at 0 (control), 0.13, 0.26, or 0.81 kg·m⁻³; see Table 1 for treatment designations. Each subplot treatment consisted of a Rootrainer (27 cells) that contained Pro-Mix BX to which appropriate amounts of slow-release fertilizer had been incorporated. On 7 Feb. 2000, single seeds were sown into each cell of the Rootrainer so that in each of the three replicate blocks, 27 seeds were sown per treatment combination. A total of 972 seeds were sown (3 blocks × 3 main plot treatments × 4 subplot treatments × 27 seeds = 972 seeds). Seedlings were watered to saturation at least once each week, or as needed during the experiment. Ten weeks after sowing, on 17 Apr. 2000, when plants attained the two- to three-leaf stage, seed germination rate was recorded. At this time 10 uniform seedlings per experimental combination, or Rootrainer, were marked in each block, and leaf number and height were determined. Weekly fertilization treatments were also initiated at this time by manually saturating media with the appropriate liquid fertilizer; water used for preparation of the treatments had a total N content [NO₃ and ammonia (NH₄)] of about 0.7 mg·L⁻¹ (Scott Laboratories, Allentown, Pa.). The number of leaves per plant and seedling height were recorded on 13 June 2000, 18 weeks after sowing, when plants were destructively harvested. Stems, leaves, and roots were oven-dried at 70 °C (158 °F) for 48 h and weighed. The greenhouse air temperature was measured every 30 min during the experiment using TEMP HOBO sensors (Onset Computer Corp., Pocasset, Mass.). Average greenhouse air temperatures (±SE) were 25.1 ± 0.1, 31.3 ± 0.3, and 34.0 ± 0.7 °C, (77.2 ± 0.1, 88.3 ± 0.3, and 93.2 ± 0.7 °F), at 0300, 1200, and 1600 HR, respectively.

EXPERIMENT 2. SLOW-RELEASE FERTILIZER APPLICATION ONLY. To

examine growth of container-grown pawpaw seedlings with slow-release fertilizer as the sole fertilizer source, a greenhouse experiment was conducted with six treatment levels of Osmocote at 0 (control), 0.81, 2.22, 4.43, 8.86, or 17.72 kg·m⁻³ (designated A to E; see Table 1). A replicated block exper- imental design, with three replicate blocks, was used with each block con- taining each of the six treatments ar- anged randomly in each block. Each treatment in each block consisted of a Rootrainer (27 cells) containing Pro- Mix BX to which appropriate amounts of slow-release fertilizer had been in- corporated. On 30 June 2000, single seeds were sown into each cell of the Rootrainer so that in each of the three replicate blocks, 27 seeds were sown per treatment combination. A total of 324 seeds were sown (3 blocks × 4 treatments × 27 seeds = 324 seeds sown). Seedlings were watered at least once each week to runoff, or as needed during the experiment. On 6 Sept. 2000, 10 weeks after sowing, plants in all treatments had attained at least the two- to three-leaf stage, and seed ger- mination rate was recorded. At this time, 10 uniform seedlings were marked, and leaf number and height were recorded. Thereafter, leaf num- ber and height measurements were recorded weekly. The number of leaves per plant and seedling height were recorded on 26 Oct. 2000, 17 weeks after the initial height measurements, when plants were destructively har- vested. Stems, leaves and roots were oven-dried at 70 °C (158 °F) for 48 h and weighed. The greenhouse air tem- perature was monitored as described previously. Average greenhouse air temperatures (±SE) were 23.5 ± 0.1, 29.1 ± 0.3, and 32.6 ± 0.7°C, (74.3 ± 0.1, 84.4 ± 0.3, and 90.7 ± 0.7°F), at 300, 1200, and 1600 HR, respectively,

STATISTICAL ANALYSIS. Data were subjected to GLM analysis of variance and GLM regression analysis using the statistical program Costat (CoHort Software, Monterey, Calif.). Both linear and polynomial regression models were fit to the data and significance of the models determined. If linear regression models were significant, they are reported. If both linear and quadratic regression models were significant, then the quadratic model was reported.

EXPERIMENT 1. SLOW-RELEASE AND WEEKLY LIQUID-FEED FERTILIZER APPLICATION. No significant treatment differences were observed for seed germination percentage (average germination percentage was 70%; data not shown). Analysis of variance indicated that there was a significant interaction between slow-release fertilizer rate and liquid-feed fertilizer regime main effects (P < 0.05) for plant height, root, shoot and total plant dry weight, and root:shoot ratio, but the interaction was not significant for seedling leaf number (Tables 2 and 3). Seedling leaf number showed a linear response to increasing rate of either liquid-feed Peters fertilizer or Osmocote, with the greatest leaf number at the highest treatment rates for both Peters and Osmocote (Table 2). Plant height showed a linear response to increasing rate of Peters in the control, medium, and high rates of Osmocote, with plants at 500 mg·L-1 Peters with the greatest leaf number (see columns in Table 2). Plant height within each level of Peters showed a quadratic response to increasing rates of Osmocote, with the greatest plant height at the highest treatment rates for both Osmocote and Peters (see rows in Table 2). Root dry weight showed a linear response to increasing rate to Peters only in the